当前位置:网站首页>寻找性能更优秀的动态 Getter 和 Setter 方案
寻找性能更优秀的动态 Getter 和 Setter 方案
2020-11-08 19:28:00 【Newbe36524】
反射获取 PropertyInfo 可以对对象的属性值进行读取或者写入,但是这样性能不好。所以,我们需要更快的方案。
方案说明
就是用表达式编译一个 Action<TObj,TValue>
作为 Setter,编译一个 Func<TObj,TValue>
作为 Getter。
然后把这些编译好的委托放在一个泛型类的静态字段中保存起来,需要使用的时候从这里面查找就可以了。
知识要点
- 使用表达式创建委托
- 泛型类的静态字段是每个闭合类型独立的,因此用于存储和类型相关的内容非常方便
实现代码
由于代码中混合的使用 Switch 作为字典的阴招,所以代码很长,此处不再罗列,仅给出链接:
基准测试
BenchmarkDotNet=v0.12.1, OS=Windows 10.0.19041.572 (2004/?/20H1)
Intel Xeon CPU E5-2678 v3 2.50GHz, 1 CPU, 24 logical and 12 physical cores
.NET Core SDK=5.0.100-rc.2.20479.15
[Host] : .NET Core 2.1.23 (CoreCLR 4.6.29321.03, CoreFX 4.6.29321.01), X64 RyuJIT
net461 : .NET Framework 4.8 (4.8.4250.0), X64 RyuJIT
net48 : .NET Framework 4.8 (4.8.4250.0), X64 RyuJIT
netcoreapp21 : .NET Core 2.1.23 (CoreCLR 4.6.29321.03, CoreFX 4.6.29321.01), X64 RyuJIT
netcoreapp31 : .NET Core 3.1.9 (CoreCLR 4.700.20.47201, CoreFX 4.700.20.47203), X64 RyuJIT
netcoreapp5 : .NET Core 5.0.0 (CoreCLR 5.0.20.47505, CoreFX 5.0.20.47505), X64 RyuJIT |
结论
- 使用委托明显比使用 PropertyInfo 要快,这个方案可以。
- Framework 真拉胯,Net 5 简直太强了。
- 如果属性是明确的,建议把字典中取出来的委托保存在自己的上下文,这可以明显的省去查找的消耗。
图表
从左往右分别是:直接读取属性、缓存委托、不缓存委托和使用 PropertyInfo。
数据
Getter
Method | Job | Runtime | Mean | Error | StdDev | Median | Ratio | RatioSD | Rank |
---|---|---|---|---|---|---|---|---|---|
DirectlyString | net461 | .NET 4.6.1 | 0.1636 ns | 0.0822 ns | 0.1126 ns | 0.1472 ns | ? | ? | 2 |
DirectlyInt | net461 | .NET 4.6.1 | 0.0318 ns | 0.0348 ns | 0.0342 ns | 0.0217 ns | ? | ? | 1 |
ReflectString | net461 | .NET 4.6.1 | 145.8375 ns | 2.2790 ns | 2.1317 ns | 145.6522 ns | ? | ? | 7 |
ReflectInt | net461 | .NET 4.6.1 | 172.5066 ns | 1.3206 ns | 1.1028 ns | 172.6804 ns | ? | ? | 8 |
GetterString | net461 | .NET 4.6.1 | 31.4379 ns | 0.6017 ns | 0.5334 ns | 31.6316 ns | ? | ? | 4 |
GetterInt | net461 | .NET 4.6.1 | 33.0642 ns | 0.4940 ns | 0.4380 ns | 33.0557 ns | ? | ? | 5 |
GetterObject | net461 | .NET 4.6.1 | 33.9174 ns | 0.5587 ns | 0.5226 ns | 33.7326 ns | ? | ? | 6 |
GetterCached | net461 | .NET 4.6.1 | 7.5878 ns | 0.1223 ns | 0.1144 ns | 7.5765 ns | ? | ? | 3 |
DirectlyString | net48 | .NET 4.8 | 0.0181 ns | 0.0353 ns | 0.0313 ns | 0.0043 ns | ? | ? | 1 |
DirectlyInt | net48 | .NET 4.8 | 0.0050 ns | 0.0089 ns | 0.0079 ns | 0.0000 ns | ? | ? | 1 |
ReflectString | net48 | .NET 4.8 | 143.8313 ns | 2.2501 ns | 2.1047 ns | 143.5568 ns | ? | ? | 5 |
ReflectInt | net48 | .NET 4.8 | 172.1714 ns | 1.9819 ns | 1.7569 ns | 172.3142 ns | ? | ? | 6 |
GetterString | net48 | .NET 4.8 | 31.5887 ns | 0.6310 ns | 0.5902 ns | 31.5385 ns | ? | ? | 3 |
GetterInt | net48 | .NET 4.8 | 32.7140 ns | 0.3992 ns | 0.3734 ns | 32.7343 ns | ? | ? | 4 |
GetterObject | net48 | .NET 4.8 | 33.3063 ns | 0.2069 ns | 0.1834 ns | 33.3053 ns | ? | ? | 4 |
GetterCached | net48 | .NET 4.8 | 7.5540 ns | 0.2201 ns | 0.1951 ns | 7.5069 ns | ? | ? | 2 |
DirectlyString | netcoreapp21 | .NET Core 2.1 | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.0000 ns | ? | ? | 1 |
DirectlyInt | netcoreapp21 | .NET Core 2.1 | 0.0193 ns | 0.0111 ns | 0.0104 ns | 0.0177 ns | ? | ? | 2 |
ReflectString | netcoreapp21 | .NET Core 2.1 | 110.4180 ns | 2.2159 ns | 1.8503 ns | 110.8038 ns | ? | ? | 7 |
ReflectInt | netcoreapp21 | .NET Core 2.1 | 138.9612 ns | 0.9694 ns | 0.8594 ns | 138.8217 ns | ? | ? | 8 |
GetterString | netcoreapp21 | .NET Core 2.1 | 16.8958 ns | 0.2384 ns | 0.2230 ns | 16.8103 ns | ? | ? | 4 |
GetterInt | netcoreapp21 | .NET Core 2.1 | 19.4407 ns | 0.2041 ns | 0.1809 ns | 19.4539 ns | ? | ? | 6 |
GetterObject | netcoreapp21 | .NET Core 2.1 | 18.6922 ns | 0.2700 ns | 0.2255 ns | 18.6582 ns | ? | ? | 5 |
GetterCached | netcoreapp21 | .NET Core 2.1 | 0.9299 ns | 0.0457 ns | 0.0427 ns | 0.9308 ns | ? | ? | 3 |
DirectlyString | netcoreapp31 | .NET Core 3.1 | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.0000 ns | ? | ? | 1 |
DirectlyInt | netcoreapp31 | .NET Core 3.1 | 0.0693 ns | 0.0102 ns | 0.0091 ns | 0.0709 ns | ? | ? | 2 |
ReflectString | netcoreapp31 | .NET Core 3.1 | 98.6735 ns | 0.8335 ns | 0.7389 ns | 98.5319 ns | ? | ? | 7 |
ReflectInt | netcoreapp31 | .NET Core 3.1 | 130.6941 ns | 0.9332 ns | 0.8730 ns | 130.5376 ns | ? | ? | 8 |
GetterString | netcoreapp31 | .NET Core 3.1 | 14.8915 ns | 0.2025 ns | 0.1795 ns | 14.8911 ns | ? | ? | 4 |
GetterInt | netcoreapp31 | .NET Core 3.1 | 16.2874 ns | 0.0789 ns | 0.0700 ns | 16.2753 ns | ? | ? | 5 |
GetterObject | netcoreapp31 | .NET Core 3.1 | 17.6202 ns | 0.1130 ns | 0.1057 ns | 17.6092 ns | ? | ? | 6 |
GetterCached | netcoreapp31 | .NET Core 3.1 | 0.6351 ns | 0.0244 ns | 0.0217 ns | 0.6393 ns | ? | ? | 3 |
DirectlyString | netcoreapp5 | .NET Core 5.0 | 0.5098 ns | 0.0328 ns | 0.0291 ns | 0.5131 ns | 1.000 | 0.00 | 2 |
DirectlyInt | netcoreapp5 | .NET Core 5.0 | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.000 | 0.00 | 1 |
ReflectString | netcoreapp5 | .NET Core 5.0 | 88.8937 ns | 0.9697 ns | 0.8596 ns | 88.7457 ns | 174.838 | 9.26 | 7 |
ReflectInt | netcoreapp5 | .NET Core 5.0 | 123.4464 ns | 1.0582 ns | 0.9898 ns | 123.3193 ns | 242.996 | 14.00 | 8 |
GetterString | netcoreapp5 | .NET Core 5.0 | 7.6628 ns | 0.0931 ns | 0.0777 ns | 7.6703 ns | 15.031 | 0.95 | 5 |
GetterInt | netcoreapp5 | .NET Core 5.0 | 6.6645 ns | 0.0825 ns | 0.0772 ns | 6.6497 ns | 13.085 | 0.69 | 4 |
GetterObject | netcoreapp5 | .NET Core 5.0 | 8.3090 ns | 0.1685 ns | 0.1576 ns | 8.2865 ns | 16.344 | 0.83 | 6 |
GetterCached | netcoreapp5 | .NET Core 5.0 | 0.9791 ns | 0.0293 ns | 0.0245 ns | 0.9764 ns | 1.920 | 0.13 | 3 |
Setter
Method | Job | Runtime | Mean | Error | StdDev | Median | Ratio | RatioSD | Rank |
---|---|---|---|---|---|---|---|---|---|
DirectlyString | net461 | .NET 4.6.1 | 2.0161 ns | 0.0300 ns | 0.0266 ns | 2.0045 ns | 1.000 | 0.00 | 2 |
DirectlyInt | net461 | .NET 4.6.1 | 0.0076 ns | 0.0094 ns | 0.0083 ns | 0.0081 ns | 0.004 | 0.00 | 1 |
ReflectString | net461 | .NET 4.6.1 | 237.5006 ns | 4.5706 ns | 4.4890 ns | 236.5912 ns | 117.871 | 3.40 | 5 |
ReflectInt | net461 | .NET 4.6.1 | 249.3627 ns | 2.1717 ns | 2.0314 ns | 249.0283 ns | 123.681 | 1.94 | 6 |
GetterString | net461 | .NET 4.6.1 | 32.8621 ns | 0.2855 ns | 0.2229 ns | 32.9189 ns | 16.335 | 0.22 | 4 |
GetterInt | net461 | .NET 4.6.1 | 33.6103 ns | 0.4245 ns | 0.3544 ns | 33.5499 ns | 16.695 | 0.26 | 4 |
GetterObject | net461 | .NET 4.6.1 | 33.2561 ns | 0.2966 ns | 0.2629 ns | 33.1795 ns | 16.497 | 0.17 | 4 |
GetterCached | net461 | .NET 4.6.1 | 9.1805 ns | 0.0761 ns | 0.0674 ns | 9.1802 ns | 4.554 | 0.08 | 3 |
DirectlyString | net48 | .NET 4.8 | 1.9272 ns | 0.0298 ns | 0.0264 ns | 1.9245 ns | 1.000 | 0.00 | 2 |
DirectlyInt | net48 | .NET 4.8 | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.000 | 0.00 | 1 |
ReflectString | net48 | .NET 4.8 | 237.7686 ns | 4.6597 ns | 5.3661 ns | 235.8445 ns | 123.908 | 3.57 | 5 |
ReflectInt | net48 | .NET 4.8 | 249.6291 ns | 4.5333 ns | 4.2404 ns | 249.5459 ns | 129.689 | 3.13 | 6 |
GetterString | net48 | .NET 4.8 | 32.2366 ns | 0.1941 ns | 0.1721 ns | 32.1780 ns | 16.731 | 0.29 | 4 |
GetterInt | net48 | .NET 4.8 | 32.0081 ns | 0.1488 ns | 0.1162 ns | 32.0270 ns | 16.572 | 0.23 | 4 |
GetterObject | net48 | .NET 4.8 | 32.6413 ns | 0.1417 ns | 0.1183 ns | 32.6260 ns | 16.907 | 0.24 | 4 |
GetterCached | net48 | .NET 4.8 | 9.2589 ns | 0.0928 ns | 0.0868 ns | 9.2564 ns | 4.799 | 0.07 | 3 |
DirectlyString | netcoreapp21 | .NET Core 2.1 | 2.4107 ns | 0.0507 ns | 0.0475 ns | 2.3936 ns | 1.000 | 0.00 | 2 |
DirectlyInt | netcoreapp21 | .NET Core 2.1 | 0.0007 ns | 0.0028 ns | 0.0025 ns | 0.0000 ns | 0.000 | 0.00 | 1 |
ReflectString | netcoreapp21 | .NET Core 2.1 | 203.6637 ns | 3.6109 ns | 3.3777 ns | 203.4793 ns | 84.517 | 2.31 | 7 |
ReflectInt | netcoreapp21 | .NET Core 2.1 | 213.8619 ns | 2.1882 ns | 1.9398 ns | 213.7367 ns | 88.757 | 1.71 | 8 |
GetterString | netcoreapp21 | .NET Core 2.1 | 19.5240 ns | 0.0811 ns | 0.0758 ns | 19.5149 ns | 8.102 | 0.15 | 5 |
GetterInt | netcoreapp21 | .NET Core 2.1 | 18.8794 ns | 0.1193 ns | 0.1058 ns | 18.8837 ns | 7.836 | 0.18 | 4 |
GetterObject | netcoreapp21 | .NET Core 2.1 | 20.6765 ns | 0.2709 ns | 0.2115 ns | 20.6419 ns | 8.584 | 0.14 | 6 |
GetterCached | netcoreapp21 | .NET Core 2.1 | 2.7606 ns | 0.0613 ns | 0.0512 ns | 2.7590 ns | 1.148 | 0.02 | 3 |
DirectlyString | netcoreapp31 | .NET Core 3.1 | 2.2625 ns | 0.0647 ns | 0.0605 ns | 2.2555 ns | 1.000 | 0.00 | 2 |
DirectlyInt | netcoreapp31 | .NET Core 3.1 | 0.0072 ns | 0.0165 ns | 0.0146 ns | 0.0000 ns | 0.003 | 0.01 | 1 |
ReflectString | netcoreapp31 | .NET Core 3.1 | 182.7306 ns | 3.3249 ns | 3.1101 ns | 182.3062 ns | 80.804 | 1.99 | 7 |
ReflectInt | netcoreapp31 | .NET Core 3.1 | 192.2510 ns | 2.3691 ns | 2.1002 ns | 191.4821 ns | 85.061 | 2.88 | 8 |
GetterString | netcoreapp31 | .NET Core 3.1 | 16.9918 ns | 0.2115 ns | 0.1979 ns | 16.9651 ns | 7.516 | 0.25 | 5 |
GetterInt | netcoreapp31 | .NET Core 3.1 | 16.1168 ns | 0.3558 ns | 0.3654 ns | 15.9822 ns | 7.138 | 0.19 | 4 |
GetterObject | netcoreapp31 | .NET Core 3.1 | 19.3060 ns | 0.4173 ns | 0.5571 ns | 19.4856 ns | 8.480 | 0.45 | 6 |
GetterCached | netcoreapp31 | .NET Core 3.1 | 2.9276 ns | 0.0156 ns | 0.0146 ns | 2.9313 ns | 1.295 | 0.03 | 3 |
DirectlyString | netcoreapp5 | .NET Core 5.0 | 2.2455 ns | 0.0084 ns | 0.0078 ns | 2.2460 ns | 1.000 | 0.00 | 2 |
DirectlyInt | netcoreapp5 | .NET Core 5.0 | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.0000 ns | 0.000 | 0.00 | 1 |
ReflectString | netcoreapp5 | .NET Core 5.0 | 162.8780 ns | 0.7135 ns | 0.6674 ns | 162.8741 ns | 72.538 | 0.45 | 7 |
ReflectInt | netcoreapp5 | .NET Core 5.0 | 171.1380 ns | 0.4173 ns | 0.3699 ns | 171.1414 ns | 76.217 | 0.31 | 8 |
GetterString | netcoreapp5 | .NET Core 5.0 | 8.6244 ns | 0.1891 ns | 0.1769 ns | 8.5469 ns | 3.841 | 0.08 | 5 |
GetterInt | netcoreapp5 | .NET Core 5.0 | 6.5511 ns | 0.0347 ns | 0.0325 ns | 6.5634 ns | 2.917 | 0.02 | 4 |
GetterObject | netcoreapp5 | .NET Core 5.0 | 9.0732 ns | 0.0306 ns | 0.0272 ns | 9.0735 ns | 4.041 | 0.02 | 6 |
GetterCached | netcoreapp5 | .NET Core 5.0 | 2.8223 ns | 0.0728 ns | 0.0681 ns | 2.8190 ns | 1.257 | 0.03 | 3 |
总结
使用表达式创建委托来取代 PropertyInfo 读取和写入属性效果很好。
开发者也可以直接引用 Newbe.ObjectVisitor 包来使用已经封装好的 ValueGetter 和 ValueSetter。
我只是知识的搬运工
- 晚绑定场景下对象属性赋值和取值可以不需要 PropertyInfo
- 三种属性操作性能比较:PropertyInfo + Expression Tree + Delegate.CreateDelegate
- 关于 Expression Tree 和 IL Emit 的所谓的” 性能差别”
发布说明
使用样例
番外分享
GitHub 项目地址:https://github.com/newbe36524/Newbe.ObjectVisitor
Gitee 项目地址:https://gitee.com/yks/Newbe.ObjectVisitor
- 本文作者: newbe36524
- 本文链接: https://www.newbe.pro/Newbe.ObjectVisitor/Better-Performance-Getter-Setter/
- 版权声明: 本博客所有文章除特别声明外,均采用 BY-NC-SA 许可协议。转载请注明出处!
版权声明
本文为[Newbe36524]所创,转载请带上原文链接,感谢
https://my.oschina.net/newbe36524/blog/4708320
边栏推荐
- SQL 速查
- Interesting article sharing: what is the difference between C language and C + +, C?
- PAT_ Grade A_ 1056 Mice and Rice
- 学会了volatile,你变心了,我看到了
- 线程池运用不当的一次线上事故
- Why need to use API management platform
- Using k3s to create local development cluster
- CountDownLatch 瞬间炸裂!同基于 AQS,凭什么 CyclicBarrier 可以这么秀?
- How does the system response time and throughput change with the increase of concurrency pressure during performance pressure testing
- Hello world of rabbitmq
猜你喜欢
abp(net core)+easyui+efcore实现仓储管理系统——出库管理之五(五十四)
How much faster is a server equipped with a SSD than a mechanical hard disk
Wechat applet related
我用 Python 找出了删除我微信的所有人并将他们自动化删除了
Brief VIM training strategy
Not a programmer, code can't be too ugly! The official writing standard of Python: pep8 everyone should know
(O)ServiceManager分析(一)之BinderInternal.getContextObject
Using GaN based oversampling technique to improve the accuracy of model for mortality prediction of unbalanced covid-19
Express框架
实验
随机推荐
Proficient in high concurrency and multithreading, but can't use ThreadLocal?
VirtualBox install centos7
Summary of interface test case ideas
精通高并发与多线程,却不会用ThreadLocal?
Dynamic ReLU:微软推出提点神器,可能是最好的ReLU改进 | ECCV 2020
Iptables from introduction to mastery
线程池运用不当的一次线上事故
接口测试用例思路总结
单例模式的五种设计方案
read文件一个字节实际会发生多大的磁盘IO?
Implementation of warehouse management system with ABP (net core) + easyUI + efcore
Mongodb add delete modify query operation
RSA asymmetric encryption algorithm
中缀表达式转后缀表达式
Tencent: Although Ali's Taichung is good, it is not omnipotent!
Dynamic relu: Microsoft's refreshing device may be the best relu improvement | ECCV 2020
趣文分享:C 语言和 C++、C# 的区别在什么地方?
experiment
Summary of rendering of water wave and caustics (etching) in webgl
Framework - SPI four modes + general device driver implementation - source code